Module org.nasdanika.ai

Package org.nasdanika.ai

Class AbstractDoubleFitter

java.lang.Object

org.nasdanika.ai.AbstractDoubleFitter

All Implemented Interfaces:

FittedPredictor.Fitter<double[],double[],Double>

Direct Known Subclasses:

AbstractMapReduceDoubleFitter, AbstractRecursiveDoubleFitter

public abstract class AbstractDoubleFitter
extends Object
implements FittedPredictor.Fitter<double[],double[],Double>

Collects all samples to double[][] features and double[][] labels, then calls fit(double[][] features, double[][] labels) to fit Function<double[][],double[][]> which makes predictions for multiple features at once. This allows to use a single base class for multiple types of predictors

Constructor Summary

Constructors

Constructor	Description
AbstractDoubleFitter()

Method Summary

Modifier and Type	Method	Description
FittedPredictor.Fitter<double[],double[],Double>	compose(FittedPredictor.Fitter<double[],double[],Double> other)
protected static Double	computeError(Function<double[][],double[][]> predictor, double[][] features, double[][] labels)
protected FittedPredictor<double[],double[],Double>	createPredictor(Function<double[][],double[][]> predictor, double[][] features, double[][] labels)
protected abstract Function<double[][],double[][]>	fit(double[][] features, double[][] labels)
<S> FittedPredictor<double[],double[],Double>	fit(Collection<S> samples, Function<S,double[]> featureMapper, Function<S,double[]> labelMapper)	Creates a predictor by fitting a collection of samples.
<S> reactor.core.publisher.Mono<FittedPredictor<double[],double[],Double>>	fitAsync(reactor.core.publisher.Flux<S> samples, Function<S,reactor.core.publisher.Mono<double[]>> featureMapper, Function<S,reactor.core.publisher.Mono<double[]>> labelMapper)	Creates a predictor by fitting a flux of samples.
static Function<double[][],double[]>	wrap(Function<double[],Double> predictor)

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Methods inherited from interface org.nasdanika.ai.FittedPredictor.Fitter

adaptFeature, adaptFeatureAsync, adaptLabel, adaptLabelAsync, compose

Constructor Details

AbstractDoubleFitter

public AbstractDoubleFitter()

Method Details

fitAsync

public <S>
reactor.core.publisher.Mono<FittedPredictor<double[],double[],Double>> fitAsync(reactor.core.publisher.Flux<S> samples,
 Function<S,reactor.core.publisher.Mono<double[]>> featureMapper,
 Function<S,reactor.core.publisher.Mono<double[]>> labelMapper)

Description copied from interface: FittedPredictor.Fitter

Creates a predictor by fitting a flux of samples.

Specified by:

fitAsync in interface FittedPredictor.Fitter<double[],double[],Double>

Type Parameters:

S -

Parameters:

samples -

featureMapper -

labelMapper -

Returns:

A mono providing a predictor. The mono can publish a predictor before the flux is finished and update the predictor with new samples from the flux.

fit

public <S>
FittedPredictor<double[],double[],Double> fit(Collection<S> samples,
 Function<S,double[]> featureMapper,
 Function<S,double[]> labelMapper)

Description copied from interface: FittedPredictor.Fitter

Creates a predictor by fitting a collection of samples.

Specified by:

fit in interface FittedPredictor.Fitter<double[],double[],Double>

Type Parameters:

S -

Parameters:

samples -

featureMapper -

labelMapper -

Returns:

fit

protected abstract Function<double[][],double[][]> fit(double[][] features,
 double[][] labels)

createPredictor

protected FittedPredictor<double[],double[],Double> createPredictor(Function<double[][],double[][]> predictor,
 double[][] features,
 double[][] labels)

computeError

protected static Double computeError(Function<double[][],double[][]> predictor,
 double[][] features,
 double[][] labels)

wrap

public static Function<double[][],double[]> wrap(Function<double[],Double> predictor)

compose

public FittedPredictor.Fitter<double[],double[],Double> compose(FittedPredictor.Fitter<double[],double[],Double> other)